Conventional MRI texture analysis of peritumoral edema in the differential diagnosis of glioblastoma and solitary metastatic brain tumor / 中华放射学杂志

ABSTRACT

Objective To investigate the value of conventional MR texture analysis of peritumoral edema in differentiating between glioblastoma and solitary metastatic brain tumor. Methods Forty patients with brain glioma and forty patients with brain metastases confirmed by pathology or clinic were enrolled . All patients underwent conventional MRI scan and enhanced examination, including axial T1WI, T2WI, fluid attenuated inversion recovery(FLAIR) and enhanced T1WI. The ROI was manually outlined by MaZda software in the most obvious level of 4 sequences of peritumoral edema and the texture features were extracted, including mean, variance, skewness, kurtosis, 1 percentile, 10 percentile, 50 percentile, 90 percentile, 99 percentile. Independent sample t-test (in case of normality and variance homogeneity test) or rank sum test (in case of not satisfying normality and homogeneity of variance) was used to compare the differences of the histogram parameters and to make meaningful parameters based on logistic regression modeling and drawing the receiver operating characteristic curve (ROC curve) to evaluate the differential diagnosis efficacy. Results Through the preliminary feature screening, skew of FLAIR (SkewFLAIR) for glioblastomas was -0.358±0.432, respectively,while for solitary metastatic brain tumor, the statistics was-0.648 ± 0.503. The results showed statistically significant difference(t=7.574,P=0.006). The differencesbetween the kurtosis (KurtFLAIR) were also proven to be statistically significant (Z=4.360 ,P=0.037),with-0.384 (-0.798, 0.049) for glioblastomas and-0.089 (-0.456, 0.613) for solitary metastasis brain tumor. The sensitivity of the corresponding logistic regression differential diagnosis model was 77.5％, specificity was 94.3％, and area under the curve (AUC) was 0.85. Conclusion Conventional MRI texture analysis of peritumoral edema provides reliable and quantified objective basis for the differential diagnosis of glioblastoma and solitary metastasis.